- Blind mode tutorial
lichess.org
Donate

ChessAgine MCP

#2

Hmm ... chessbase will use it and claim they created it and sell it for 200€

Hmm ... chessbase will use it and claim they created it and sell it for 200€
#3

@kofokofoka said in #2:

Hmm ... chessbase will use it and claim they created it and sell it for 200€

even if they do, people won't pay for something that is free ;))

@kofokofoka said in #2: > Hmm ... chessbase will use it and claim they created it and sell it for 200€ even if they do, people won't pay for something that is free ;))
#4

Hey, I would like to contribute, may I know how you got this idea, what sources you have surfed for this...

Hey, I would like to contribute, may I know how you got this idea, what sources you have surfed for this...
#5

@Noobmasterplayer123 said in #3:

Hmm ... chessbase will use it and claim they created it and sell it for 200€

even if they do, people won't pay for something that is free ;))

well, people bought the fatfritz even if it was stockfish in disguise ...

@Noobmasterplayer123 said in #3: > > Hmm ... chessbase will use it and claim they created it and sell it for 200€ > > even if they do, people won't pay for something that is free ;)) well, people bought the fatfritz even if it was stockfish in disguise ...
#6

@finday said in #5:

Hmm ... chessbase will use it and claim they created it and sell it for 200€

even if they do, people won't pay for something that is free ;))

well, people bought the fatfritz even if it was stockfish in disguise ...

Yeah, I was kidding, the project is licensed under GPL, and the extension has an author field. If they were to fork it, they would have to open source it

@finday said in #5: > > > Hmm ... chessbase will use it and claim they created it and sell it for 200€ > > > > even if they do, people won't pay for something that is free ;)) > > well, people bought the fatfritz even if it was stockfish in disguise ... Yeah, I was kidding, the project is licensed under GPL, and the extension has an author field. If they were to fork it, they would have to open source it
#7

@CupidKnightRider said in #4:

Hey, I would like to contribute, may I know how you got this idea, what sources you have surfed for this...

Well, I have been trying to use LLMs for analysis, so just following normal MCP protocol from anthropic, I just created a ChessAgine MCP. The PTAG tools are my idea to break eval into numbers. Code is os, feel free to join the community, there are some ideas you can work on also.

@CupidKnightRider said in #4: > Hey, I would like to contribute, may I know how you got this idea, what sources you have surfed for this... Well, I have been trying to use LLMs for analysis, so just following normal MCP protocol from anthropic, I just created a ChessAgine MCP. The PTAG tools are my idea to break eval into numbers. Code is os, feel free to join the community, there are some ideas you can work on also.
#8

@Noobmasterplayer123 said in #7:

Hey, I would like to contribute, may I know how you got this idea, what sources you have surfed for this...

Well, I have been trying to use LLMs for analysis, so just following normal MCP protocol from anthropic, I just created a ChessAgine MCP which brings all ChessAgine features to Claude desktop. The PTAG tools are my idea to break eval into numbers. Code is os, feel free to join the community, there are some ideas you can work on also.

@Noobmasterplayer123 said in #7: > > Hey, I would like to contribute, may I know how you got this idea, what sources you have surfed for this... > > Well, I have been trying to use LLMs for analysis, so just following normal MCP protocol from anthropic, I just created a ChessAgine MCP which brings all ChessAgine features to Claude desktop. The PTAG tools are my idea to break eval into numbers. Code is os, feel free to join the community, there are some ideas you can work on also.
#9

cool but is claude really doing anything useful here? it seems to just be adding unnecessary verbiage to the puzzle-solving experience. look at how lichess already implements chess puzzles- it prompts the user for a move at some position and provides a hint button that highlights the correct piece. So how is that different from hint #4? and then the rest of the generated text is the standard boilerplate for solving puzzles that any club player would know intuitively. As it stands right now I dont see much value in what you built.

What I think would be interesting is if you combine stockfish and claude to give a context-aware description of some position, describing typical plans or tactical motifs that either player could implement. essentially, would it be possible for these two parties to describe a position in the same way that a grandmaster coach would to a student? is that feasible given the architecture youve build so far?

cool but is claude really doing anything useful here? it seems to just be adding unnecessary verbiage to the puzzle-solving experience. look at how lichess already implements chess puzzles- it prompts the user for a move at some position and provides a hint button that highlights the correct piece. So how is that different from hint #4? and then the rest of the generated text is the standard boilerplate for solving puzzles that any club player would know intuitively. As it stands right now I dont see much value in what you built. What I think would be interesting is if you combine stockfish and claude to give a context-aware description of some position, describing typical plans or tactical motifs that either player could implement. essentially, would it be possible for these two parties to describe a position in the same way that a grandmaster coach would to a student? is that feasible given the architecture youve build so far?
#10

@Presiident said in #9:

cool but is claude really doing anything useful here? it seems to just be adding unnecessary verbiage to the puzzle-solving experience. look at how lichess already implements chess puzzles- it prompts the user for a move at some position and provides a hint button that highlights the correct piece. So how is that different from hint #4? and then the rest of the generated text is the standard boilerplate for solving puzzles that any club player would know intuitively. As it stands right now I dont see much value in what you built.

What I think would be interesting is if you combine stockfish and claude to give a context-aware description of some position, describing typical plans or tactical motifs that either player could implement. essentially, would it be possible for these two parties to describe a position in the same way that a grandmaster coach would to a student? is that feasible given the architecture youve build so far?

The puzzle session is just 1 of the use cases, and for your point of tactical motifs PTAG tool + Claude can identify thematic elements of a position and put it in words. It can even give you an eval test where it can get puzzles from Lichess and ask you to tell positional elements, and it will verify it with Stockfish, Puzzle Database and even master DB for the move you give and your explanation and tell you where you went wrong and what could be improved. As I said in the blog, clearly it's not at the GM level, it's a tool that can be used for many tasks, puzzle session is one of many things this integration can do. You are judging a book by its cover, and nothing I can do about it!

@Presiident said in #9: > cool but is claude really doing anything useful here? it seems to just be adding unnecessary verbiage to the puzzle-solving experience. look at how lichess already implements chess puzzles- it prompts the user for a move at some position and provides a hint button that highlights the correct piece. So how is that different from hint #4? and then the rest of the generated text is the standard boilerplate for solving puzzles that any club player would know intuitively. As it stands right now I dont see much value in what you built. > > What I think would be interesting is if you combine stockfish and claude to give a context-aware description of some position, describing typical plans or tactical motifs that either player could implement. essentially, would it be possible for these two parties to describe a position in the same way that a grandmaster coach would to a student? is that feasible given the architecture youve build so far? The puzzle session is just 1 of the use cases, and for your point of tactical motifs PTAG tool + Claude can identify thematic elements of a position and put it in words. It can even give you an eval test where it can get puzzles from Lichess and ask you to tell positional elements, and it will verify it with Stockfish, Puzzle Database and even master DB for the move you give and your explanation and tell you where you went wrong and what could be improved. As I said in the blog, clearly it's not at the GM level, it's a tool that can be used for many tasks, puzzle session is one of many things this integration can do. You are judging a book by its cover, and nothing I can do about it!